VCGAN: Video Colorization with Hybrid Generative Adversarial Network

We propose a hybrid recurrent Video Colorization with Hybrid Generative Adversarial Network (VCGAN), an improved approach to video colorization using end-to-end learning. The VCGAN addresses two prevalent issues in the domain: Temporal consistency and unification of network refinement into single architecture. To enhance quality spatiotemporal consistency, mainstream generator is assisted by additional networks, i.e., global feature extractor placeholder extractor, respectively. encodes semantics grayscale input quality, whereas acts as feedback connection encode previous colorized frame order maintain consistency. If changing for input, also has potential perform image colorization. improve far frames, we dense long-term loss that smooths temporal disparity every remote frames. Trained losses jointly, strikes good balance between color vividness continuity. Experimental results demonstrate produces higher-quality temporally more consistent colorful videos than existing approaches.